How the numbers work

Prices are TCGplayer's own daily market figures, captured once per day for every product and printing (Normal / Holofoil / Reverse Holofoil / editions). Marketis TCGplayer's trailing blend of actual sales: a fair value estimate, not an executable quote. Lowis the lowest current ask, which a single troll listing can distort on a thin book. Our archive runs back to February 2024, one snapshot per card per day, no gaps we don't heal.

Charts default to a 7-day rolling median to cut single-day noise; RAW is one tap away on every chart.

The Estimate answers one question: what would a patient buyer actually pay for this card, delivered, today? It is a buy-side number. We do not claim "what it last traded at": we do not have per-sale sold data, and pretending asks are sales is the dishonesty we built this site against.

The formula, in full, every constant included:

1. Take the live eBay delivered asks for the card's representative printing and drop outliers: anything above 2.5x or below 0.3x the ask median. The cheapest survivor is the eBay floor.
2. The venue flooris the cheaper of the eBay floor and TCGplayer's lowest ask: the cheapest real, deliverable price anywhere we track.
3. The Estimate is TCGplayer's market blend, clamped between the venue floor and the eBay floor: trusted while it sits inside executable reality, overridden when it drifts outside it (the classic mid-era and vintage failure mode). With only one source available, the Estimate is that source, labeled as such.

Every Estimate carries a grade: SOLID (eBay scan at most 7 days old, at least 5 surviving asks, sources within 20%), FAIR (scan at most 14 days, at least 2 asks, within 50%), or THIN(anything less). THIN means the market's data is too shallow to corroborate the number yet: a single source, sparse or stale asks, or a solo-variant reprice. It grades the depth of the evidence, never the honesty of the arithmetic, and always comes with the reason in words. The SHOW THE WORK panel on every card page evaluates the formula in front of you and links each input to the venue that produced it. A THIN estimate is never cited in our published content.

Changelog: v1 (July 2026), first release. Any change to the formula or constants bumps the version, here and on every card page.

A MINT ALERT marks a move that is statistically unusual for that specific card: the 7-day move exceeds 2 standard deviations of the card's own daily volatility, measured over the 90 days before the move began. It grades the move against the card's own history, never against a model of the future. There is no forecast anywhere on this site.

The constants, in full:

1. Volatility = standard deviation of daily log returns over the 90 days ending where the move window starts, scaled to the window (sigma × √7 for the weekly boards).
2. Alert threshold: |move| > 2 × window volatility. A minimum of 30 daily observations is required; a card we cannot grade is never flagged.
3. A daily volatility floor of 0.2% stops dead-flat tapes from producing absurd scores, and moves under 10% are never alerts regardless of the math: an alert must be unusual AND material.
4. The frequency line ("moves this big happen about twice a year for this card") is the two-tailed normal expectation of a same-or-bigger weekly move, times 52 weeks. It describes the card's past tape, not a prediction.

Thin-market and solo-variant flagged moves are excluded before grading, and drawdown context ("12% below its March 2026 high") comes straight from the same daily archive.

The same formula over a 30-day move windowwith a higher bar: |move| > 3 standard deviations of the card's own 90-day tape (sigma × √30), same 10% materiality floor, same volatility floor, same thin/solo exclusions. Nothing else changes — an unusual month instead of an unusual week.

The ledger records the 10 strongest alerts per direction per day, ranked by how many sigmas the move cleared. Every recorded alert is scored at 7, 30 and 90 days on the public scoreboard, same as every other lane — wins and losses alike, no deletions.

Why these constants: we replayed the formula over the full daily archive (2.4 years, every card) across a grid of windows and thresholds, against the market's own drift and a random placebo, before choosing them. The scoreboard, not the backtest, is the track record: past behavior of the tape, never a forecast.

⚠ THIN: very few listings or sales back this move; the price may hinge on a single stale or troll listing. Treat with suspicion.

⚠ SOLO: only this printing of the card moved while its sibling variants were flat. Often a repricing artifact rather than real demand: real demand lifts every printing of a card.

We flag rather than hide: an honest board with warnings beats a quiet board that lies by omission.

Every "net" figure subtracts real selling costs, the number you would actually bank, not the sticker:

eBay net       = price − 13.25% final value fee (2.35% on the part over $7,500)
                 − $0.30/order ($0.40 on orders over $10) − shipping
TCGplayer net  = price − 10.75% commission (capped at $75)
                 − 2.5% + $0.30 processing − shipping
shipping       = $1.00 plain-white-envelope default for single cards

Rates are the common single-card seller case (TCGplayer Level 1 to 4 marketplace; eBay trading-cards category), not every edge case. Promotional discounts like eBay's 50%-off on $1,000+ singles are opt-in and not assumed. Fee schedules last verified 2026-07-07.